Process-based therapy: A personalized approach to treatment

Abstract

Historically, evidence-based treatment has followed the latent disease model, which emphasizes using specific protocols tied to diagnoses. Today, the field continues to move towards an individual approach with models of treatment based on change processes. Here, we describe Process-Based Therapy (PBT), a new way of thinking that is moving away from nomothetic studies of diagnosis-driven interventions toward an individual approach to treatment that recognizes the complexity of human suffering. In PBT, therapists select from a wide range of evidence-based interventions, tailoring treatment to meet a person’s needs at a given point in time. PBT is used to analyze intra-individual changes at the level of complex networks of biopsychosocial events, then gathering these into subpopulation and overall population parameters using theory and experimental analysis. PBT emphasizes tracking patient progress over time and treating symptoms based on current experiences, as well as understanding a patient’s past and predicting future experiences. Through specific analyses that aid in this process, therapists can use PBT to create a network with clients to visualize symptoms over time and areas of change.

Evidence-based treatment in clinical psychology has traditionally adhered to the latent disease model, conceptualizing psychological problems as caused by underlying latent diseases with specific biological dysfunctions that are yet to be uncovered. The Diagnostic and Statistical Manual of Mental Disorders (DSM) and International Statistical Classification of Diseases and Related Health Problems (ICD) are still firmly rooted in this model of pathology. Another approach adopted by early behavioral therapy examined functional and contextual relationships to explain and modify behaviors. With the cognitive revolution in the 1980s, behavioral approaches merged with cognitive models to what has become known as cognitive behavioral therapy (CBT). CBT has since become the gold-standard psychological approach to psychological disorders, highlighting the relationship among our cognitions—especially automatic thoughts—behaviors, and emotions (Wright & Beck, 1983). Over the years, specific cognitive models have been developed for DSM and ICD-defined disorders. This approach to treatment of psychological disorders has been to use protocols tailored to specific DSM diagnoses (e.g., generalized anxiety disorder, obsessive-compulsive disorder), with protocols validated in randomized controlled trials (RCTs). This means of establishing protocol efficacy has dominated clinical psychology research for decades (e.g., Chambless & Hollon, 1998; Tolin, 2020; Tolin et al., 2015).

Several issues have arisen, however, with these nomothetic approaches to treatment. High comorbidity exists between psychological disorders (Brown et al., 2001), which puts into question whether one can accurately follow diagnostic-based protocols without neglecting other potentially pertinent aspects of clients’ clinical presentations. In addition, individual differences cannot be accounted for with one standard approach. The field is thus moving towards developing and implementing individualized treatment, and in line with this is a call to focus on biopsychosocial change processes (Hofmann & Hayes, 2019). An example of the shift toward this process is the National Institute of Mental Health (NIMH) Research Domain Criteria (RDoC) program, which examines mental disorders based on psychological, behavioral, and neurobiological measures and processes, rather than just symptoms alone (Insel et al., 2010). Another example of this shift is the network approach, which links biological, psychological and social processes to networks of individual symptoms to address qualitative similarity in network structure across disorders (Curtiss & Klemanski, 2016).

The deeper rationale behind this movement toward process is the growing recognition that group-level analyses—as is used in most randomized controlled trials—do not apply to the developmental trajectories of individuals (Molenaar, 2004). As a result, the field is beginning to move away from a nomothetic model of treatment that treats individual differences as statistical noise and toward an idionomic (portmanteau of “nomothetic” and “idiographic”) model based on a coherent set of evidence-based change processes modeled at an idiographic or individual level, then extended to nomothetic generalizations without distorting findings at the individual level.

We argue that this shift towards more personalized treatment is necessary, and this shift comports with the development of process-based therapy (PBT), a meta-model of evidence-based therapy representing a new way to think about existing forms of evidence-based intervention (Hofmann & Hayes, 2019). In essence, PBT changes the key question in clinical psychology from “What treatment packages best improve outcomes for particular disorders?” to “What core biopsychosocial processes should be targeted with this client given this goal in this situation, and how can they most efficiently and effectively be changed?” (Hofmann & Hayes, 2019, p. 2). The overarching goal of PBT is to decrease human suffering and promote prosperity. PBT involves identifying (1) which core biopsychological processes to target in an individual given their specific goals, context, and stage of intervention and (2) how to most effectively do so using idionomic analysis, complex network approaches, and experimental psychopathology (EPP), implementing evidence-based intervention strategies.

PBT focuses on functional themes that explain the mental, social, or behavioral health difficulties experienced by the client (e.g., avoidance of fear of rejection) over syndromal classification that is put atop individuals but fails to explain within-person life trajectories (e.g., depression, anxiety). This is important, as individuals do not merely exist in a vacuum, but are influenced by their past and current context. Conceptualizing clients using normative diagnostic labels (e.g., “Client A has major depression, Client B has major depression and social anxiety disorder”) at best leads to a paucivariate additive view of human complexity instead of the known interactive nature of developmental history, cultural identity, current context, recent stressors, interpersonal style, and functional role of “symptoms” (Fried & Nesse, 2015). Additionally, even individuals with similar symptoms often experience very different life trajectories (Menzies et al., 2021). Thus, in contrast, PBT focuses on person-specific variables that are theory-based, dynamic, experimental, progressive, contextually bound, multi-level, and have been empirically established to link to key outcomes. PBT incorporates similar theories to EPP, which aims to uncover mechanisms underlying variables and processes that contribute to or maintain psychopathology (Forsyth & Zvolensky, 2001). EPP also aims to better inform treatment using experimental methods, in combination with methods such as experience sampling, neuroimaging, and psychophysiology. PBT is similar in this way, as it involves understanding at the individual level and being specific and flexible with treatment approaches. For example, PBT may entail targeting ruminative cognitive processes linked to low self-worth that drive depressed mood or worry, for a person who has a history of being bullied, rather than attempting to treat the diagnoses of “major depressive disorder” and “generalized anxiety disorder.” Thus, PBT highlights the importance of symptoms, rather than targeting distinct disorders. PBT considers the individual symptom-level and then considers plausible explanatory processes based on theory, research, and the patient.

A corollary of the shift in assessment focus from syndromes to processes of change is that treatment can in principle be more modular and draw from a wider range of theoretical orientations and methods because therapists are no longer treating diagnoses, but rather individuals living in their unique context. This means therapists can be creative and flexible in terms of which treatment strategies they use, or which skills to pull from a variety of resources, regardless of their provenance. As long as therapists use strategies that target person-specific, defined, measurable change processes linked ideographically to the client’s presentation and goals, and evaluate the effectiveness of their intervention against client’s desired outcomes, they are making an evidence-based attempt to alleviate that person’s specific struggles and enhancing well-being.

By firmly undergirding treatment in longitudinal idionomic analysis, PBT moves beyond vapid eclecticism to testable hypotheses that build on the client’s strengths and target problem areas in the service of personally chosen goals. For instance, a therapist may hypothesize that, for a particular client, targeting social isolation using behavioral activation strategies will increase frequency of interpersonal interactions and engagement in hobbies. Using high temporal density longitudinal data from that client and complex network analysis, the therapist can determine whether their hypothesis was supported or needs to be revised.

Extended evolutionary meta-model (EEMM)

PBT is grounded in evolution science principles to accommodate various evidence-based therapeutic models and theories. This is because PBT considers psychological disorders to be mal-adaptations, which evolutionarily are caused by problems in variation, selection, and/or retention of specific biopsychosocial dimensions in a given context (Hayes et al., 2020). PBT also uses the extended evolutionary meta-model (EEMM) as a means of organizing processes along psychological dimensions (affective, cognitive, attentional, self, motivational, overt behavioral), nested within other levels of analysis (biophysiological, sociocultural), in response to evolutionary variables that impact the dimensions and levels (variation, selection, retention, context) (Hayes et al., 2020). Researchers and clinicians can use the EEMM to identify, study, categorize, and target the processes related to in maladaptive functioning. The Research Domain Criteria (RDoC) can be informed by EEMM principles as well. RDoC aims to classify mental disorders based on dimensions of observable behavior and neurobiological measures (Insel et al., 2010). The RDoC initiative uses varying levels of analyses—including molecular, brain circuit, symptom, and behavioral levels—to define constructs proposed to be important symptoms of mental disorders (Moskow et al., 2022). This reflects the important shift towards processes of change the field is moving towards; the EEMM adds helpful organizational concepts drawn from evolutionary theory.

Wilson et al. (2014) explain the main facets of the EEMM in terms of variation, selection, retention, and context, and we emphasize considering and applying these concepts across multiple domains. Variation is the important first step toward adaptation (Hayes et al., 2019), as improvement requires change. The next important step is healthy selection. Examples of types of selection processes include reinforcement, values accomplishment, goal attainment, needs satisfaction, or social attachment. Healthy selection processes are important for serving adaptive functions, as they determine which variations are more likely to re-occur. Retention involves maintaining gains and creating new healthy habits that are resistant to relapse. Context includes the factors that must be considered to determine whether desired selection and retention will occur as issues such as diversity, culture, social support, existing repertoire, and family can play a large role in determining which moves are helpful in a given situation. Multilevel and multi-dimensional selection is important, which is how evolutionary factors operate across many systems such as gene systems, behavioral classes, cognitive themes, biophysiological, and sociocultural levels (Hayes et al., 2019).

Clinicians can use core change processes to determine how each of these domains interact to form maladaptive networks of thoughts, emotions, and behaviors. Therapeutic changes can also be seen as patients use these same evolutionary dimensions to form more adaptive responses through treatment (Hofmann & Hayes, 2019; Moskow et al., 2022). In the therapy room, a treatment target hierarchy can help therapists identify which problems their patient identifies as most important, and therefore which problem areas to focus on, while leveraging existing client strengths or developing new ones. Stages within a target hierarchy can provide a process to help organize how session time is allocated. The targeted content is designed to elicit change in a client in the direction of their goals and values.

An example can be seen with utilizing EEMM for a patient with social anxiety disorder (SAD). For instance, a patient may present with a negative core belief that they are worthless. Clinicians can use core change processes to determine the ways in which selection, retention, variation, and context interact to form these negative thoughts this person has about themselves (“I am stupid” or “people think I am weird”) and their impact on anxious symptoms and avoidance behaviors, such as no longer attending social events so as to avoid painful thoughts linked to the core belief. The therapist may then create a treatment exposure hierarchy, with a list of problem areas that are important to the patient, including items like calling friends, initiating conversations, and attending social events. Treatment can be constructed from processes that underly varying symptoms of psychopathology as well. For instance, common processes of poor emotion regulation and maladaptive cognitive appraisals have been suggested to relate to many different expressions of anxiety (Barlow et al., 2004).

PBT in clinical settings

Similar to adopting PBT principles in research, using PBT in clinical practice also necessitates a fundamental change in how we conceptualize and assess psychological struggles. Personalized treatment in PBT is possible following a personalized diagnosis of what is going on at the level of processes (e.g., rumination) and products (e.g., individual symptoms that may be present). Rather than reify diagnostic labels and their symptoms, PBT asks clinicians to consider how various processes are maintained and relate to one another to produce a self-sustaining pathological network of interrelated nodes or events (e.g., “low motivation,” “avoidance of public speaking”) that will form the basis of treatment planning. For clinicians who adhere staunchly to the DSM, this may appear to complicate initial assessment and case conceptualization, because in some ways, it is easier to identify a limited set of topographical symptoms (e.g., uncontrollable worry, muscle tension, insomnia) and to match symptoms to an appropriate DSM diagnosis (e.g., generalized anxiety disorder). Clinicians already know what to ask (e.g., “Do you find your worries difficult to control?”) and only need to categorize what they observe to known groups. The problem is that the idiographic fit and personalized treatment implications of this traditional approach is very limited.

In contrast, the network bottom-up approach starts from the client, not the DSM. Multi-level, multi-dimensional biopsychosocial processes of change can be targeted and an idionomic approach can be used, involving frequent idiographic assessment and scaling nomothetic findings when this improves idiographic fit (Hayes et al., 2022). Network approaches can guide this effort towards individualized treatment, by helping inform which measures and methods are best to use. The network should reflect the client’s story of their struggle, including how various elements influence each other. Thus, questions and discussion during an intake may explore how seemingly disparate aspects of a client’s struggle are interconnected and depend on context, which is a vastly divergent approach from asking yes/no questions from a list of symptoms. Admittedly, the DSM categorical approach has the advantage of facilitating communication of the severity of presenting concerns and type and amount of treatment needed with clients, patients, other healthcare providers, and insurance companies. Until the field is more comfortable with processes of change as a focus, describing each client’s presenting concern in a person-focused way may hinder a sense of understanding that is facilitated by normative concepts, even if these concepts include relatively diverse collections of individuals.

In essence, similar to most evidence-based treatments, PBT strives to identify approaches that work for everyone, but it does so through bottom-up methods (i.e., understanding the individual before searching for general patterns) as opposed to top-down or nomothetic methods (i.e., understanding the group and assuming it will generalize to the individual). An example of a statistical method that exemplifies this approach is Group Iterative Multiple Model Estimation (GIMME), which estimates individual-level networks before identifying subgroups of individual networks that share sufficiently similar patterns, retaining edges that increase idiographic fit for most individuals (Lane et al., 2019). In other words, PBT researchers and clinicians still deeply care about commonalities and pattern detection across individuals, but not at the expense of recognizing individual-level variation.

Another possible barrier—if clinicians choose to use idiographic network-based assessments—is that they will most likely need to learn how to integrate EMA into existing routine outcome monitoring procedures, which poses a logistical and statistical challenge. Typically, EMA entails prompting a client several times a day at random intervals, so clinicians would need a way to maintain contact with clients and have them provide data in between sessions. Moreover, from a PBT perspective, EMA items may be more useful if they are personalized, meaning that clinicians may need to spend effort considering which items to include in prompts for each client rather than simply issuing a standard set of self-report instruments. Subsequently, even if data are successfully collected, clinicians are faced with the task of interpreting swaths of longitudinal data, whether it be collating them into a coherent network or identifying early warning signals. For clinicians less familiar with advanced statistical methods, this is a tall order. However, eventual automation of data collection, analysis, and interpretation may make EMA less intimidating for clients and clinicians.

EPP can guide approaches to choosing which evidence-based interventions to deliver as well and the best methods to use. Clinicians who deliver evidence-based therapies already have the requisite skills to conduct assessment and intervention consistent with PBT principles. The missing piece is organizing these strategies and ideas under a theoretically coherent umbrella to inform how to apply those skills. In PBT, this overarching theoretical framework is the EEMM, which provides organization with respect to the psychological dimension being targeted, level of analysis, evolutionary stage, and context. For instance, to increase variation in the behavioral dimension, clinicians can easily draw on behavioral activation or opposite action from dialectical behavior therapy (DBT), and to improve healthy selection, they can use values or contingency management to ensure adaptive behaviors are reinforced. Relatedly, the EEMM may also encourage clinicians to consider sociocultural factors that affect client functioning (e.g., client may lack social support to engage in behavioral activation relying on group participation) and the context in which processes are relevant (e.g., behavioral activation may work for depressed mood, whereas opposite action may be more effective for anger and shame). None of these techniques are new, but the framework in which they are used is. Thus, the incremental contribution of the EEMM above and beyond network approaches is overlaying a metatheoretical framework on an otherwise theoretically agnostic methodology. After all, networks approaches are a tool; nothing in network approaches stipulates which variables should be included or excluded in the analysis—that depends on the intentions and ambitions of creator of the network. The EEMM is designed to clarify exactly those intentions and goals.

In reality, many clinicians we consider especially adept are likely already doing some version of PBT. They hold case conceptualizations specific to individual clients, understanding how learning history and current environment color the problems with which they struggle, such that there are no two clients who require the exact same treatment approach. That is because even if the problems identified between clients are similar, the contexts in which those problems occur are inherently different. Proficient clinicians are also able to flexibly deploy evidence-based techniques to match their clients’ struggles at a specific time rather than commit to a one-size-fits-all protocol. In other words, PBT is likely already being done without being labeled as such. The issue is that, without an organizing framework, measurement tools, and analytic methods that fit such a person specific approach, the case conceptualization skills of talented clinicians are exceedingly difficult to replicate and propagate. This is the clinical gap PBT is designed to fill. Through laying out the EEMM and network approach (Hofmann et al., 2021), PBT aims to codify person-specific assessment and intervention methods applied in a context-sensitive fashion, so that “clinical talent” becomes an empirical science rather than a je ne sais quoi that remains outside the reach of most clinicians.

Challenges in PBT research

Despite the wealth of knowledge gathered from clinical trials and meta-analyses of the various forms of CBT, results from nomothetic studies of diagnosis-driven interventions have limited known generalizability to individual performance. Aggregated data necessarily fail to accurately reflect the individuals they are presumed to represent because human phenomena are rarely—if ever—ergodic (i.e., all humans are the same and each human process stays the same over time), which is a little known but required assumption of classical normative statistical methods (Fisher et al., 2018; Molenaar, 2004). Although this means that therapy cannot be a one-size-fits-all endeavor, given that every person and their struggles are unique and occur in a unique context, psychological theory can bridge the gap from idiographic analysis to nomothetical generalizations through inductive reasoning and methods.

Accomplishing that end goal requires that clinical researchers take an idionomic approach to the collection and analysis of relevant data in which ecological momentary assessment (EMA) reflects life as it is lived over time by individuals, allowing each person first be “heard” (that is, to be described and fully characterized via measurement) as the foundation for characterizing treatment impact. Nomothetic generalizations can then be sought but only if they do not “drown out” the voices of the individual participants (Lane & Gates, 2017). Broadly speaking, this approach to intervention development and evaluation builds on a long tradition of idiographic functional analytic methods in behavior analysis and therapy (e.g., single-case designs) but carries it into the modern era using technological advances in data collection and analysis including EMA and complex network analysis. One challenge to implementing an EMA approach, however, is determining how long to follow an individual and measure constructs of interest to obtain a reliable person-level understanding. This is further complicated by the fact that most measures have been validated at the group-level, often with psychometric methods that themselves violate assumptions of ergodicity. EPP methods could be a way to help advance measurement, by obtaining different indicators of the same process. Further work in this area will aid in our research related to idionomic approaches.

Psychology research has long focused on treatment outcomes related to symptom reduction, to find reliable results to meaningful questions. Hofmann et al. (2020) argued that to progress our knowledge about treatment outcomes, it is essential to focus on processes of change. Functional groupings must be emphasized rather than continuing to study the many different protocols that exist for what has been thought of as divergent disorders (Weisz et al., 2018). The main method of studying processes of change in the literature has been linear mediation (treatment → mediator → outcome) of randomized controlled trials (Hofmann et al., 2020). However, several problems exist with a traditional mediation approach. For instance, a nomothetic approach does not encapsulate the entirety of therapy for an individual (Hofmann & Hayes, 2019). Thus, one approach for researchers to take is to focus on person-level or idiographic data to better understand variation occurring within the individual and how the individual is responding to treatment, rather than aggregating data across individuals. Researchers can also shift toward using a dynamic network approach that accounts for multiple variables simultaneously and bidirectional relationships among those variables (Hofmann et al., 2020). These methods can be realized using intensive longitudinal data collected from the individual via EMA and subsequently personalizing treatment based on individual-level networks (for examples, see Fisher et al., 2019; Levinson et al., 2021)

Another problem with traditional statistical approaches is that they fail to take into account changes over time or may make the assumption that relationships between two processes are linear. Hofmann et al. (2020) discussed transitioning from mediators and moderators to treatment or therapeutic processes, which PBT emphasizes. The statistical analyses involved in PBT seek to understand meaningful changes at the individual level, in consideration of context, non-linear progress that builds across time, and cyclical symptom relations. Examples of statistical approaches that may be used for process-based research include complex network analysis, time-series analysis, and statistical process control procedures to detect early warning signals related to a presenting concern (Barthel et al., 2021; Schat et al., 2021). In the context of PBT, complex network analysis entails estimating an individual-level network comprising relevant processes of change (e.g., self-critical thoughts, difficulty identifying emotions) and environmental events (e.g., recent break-up, parents were/are invalidating), which forms the basis of PBT case conceptualization. This network can be determined via clinical interview (i.e., based on clinical judgment) or empirically (i.e., based on longitudinal data), though a combination of both approaches may provide a more balanced picture (Burger et al., 2021; Ong et al., 2022).

Because PBT calls for the use of different methods to answer different questions from what has been historically ingrained in clinical psychology, getting buy-in from most clinical researchers may take time. Furthermore, current knowledge of how to conduct intervention research may need to be updated if we are to move away from the RCT horserace and toward person-centered investigations. Thus, successfully transitioning to PBT-informed research requires not just a foundational shift in scientific perspective and clinical conceptualization but also accumulation of new skills, each of which presents significant barriers to the implementation of PBT in research.

Conclusion

This article outlines the basics of PBT and highlights its growing importance in the field of clinical psychology. PBT continues to be developed through idiographic assessment and analysis (Ciarrochi et al., 2022), with the goal of understanding meaningful idionomic changes in the service of alleviating suffering and promoting prosperity. Through approaches such as EMA and network analyses, and CSD and tipping points, PBT analyzes changes on an individual level, recognizing the dynamic interplay among variables that are themselves constantly changing. In a clinical setting, PBT emphasizes tracking client processes over time and treating problems using theoretically coherent methods and falsifiable hypotheses, based on a client’s current experiences and goals. Our hope is that clinical research too will continue to prioritize studying treatment at the individual level, before making group-level inferences, through a process-based approach.

The promise of widespread implementation of PBT includes more idionomic studies that will shape the empirical landscape of evidence-based treatments for clients, increased coherence across the gamut of evidence-based modalities and orientations, more flexible treatment options with less diagnostic specialization, and more culturally informed care through explicit integration of cultural context into treatment planning. Ultimately, if these goals can be accomplished, more people will have access to treatment that is designed to be effective for them in their unique environment, not for an imaginary statistical average.